The latest data from the United States Enterprise Expenditure Management Corporation, Ramp, indicate that OpenAI is growing more rapidly among business clients and is narrowing the gap with Anthropic. Although Anthropic is still in the lead, the fact that business users switch between models suggests that this market pattern remains unstable.

Anthropic is still in the lead.

Ramp has collected data on billing and business card expenditures for over 70,000 United States enterprises. These enterprises spend billions of dollars on a combined scale, with customers covering a wide range of industries, but are more in favour of technology companies as a whole.

Pressing Ramp's caliber, Anthropic surpassed OpenAI for the first time in May this year. At that time, Anthropic ' s share of the users of the paid enterprise was 41 per cent and OpenAI 39 per cent. By July, Anthropic's share was close to 44%, OpenAI was close to 40%.

OpenAI is growing faster this quarter.

However, Ramp Economist Ara Kharazian says that if we look only at data so far in the third quarter of this year, OpenAI is growing faster than Anthropic in this client group.

Ramp did not disclose the corresponding actual expenditures of the two companies and published only the percentage data. As a result, this set of data is more appropriate to observe trends than to directly measure the real income size of the two companies.

  • May: Anthony 41%, OpenAI 39%
  • July: Anthropic approaches 44%, OpenAI approaches 40%
  • The sample covers over 70,000 American businesses.

Model update affects business selection

On the X platform, Kharazian states that OpenAI ' s new model GPT-5.6 Sol is strong and is increasingly becoming a developer ' s choice. At the same time, he indicated that the high-end model level of Anthropic Fable 5 had not performed as well as expected in terms of adoption rates and practical applications, and that the data retention requirements resulting from price and regulation could be one of the reasons.

TechCrunch mentioned that this set of data does not represent the entire enterprise AI market, as large enterprises often use other expenditure management tools, such as American Express, and not necessarily through Ramp. However, the sample still reflects a signal that there are no fixed stations for business clients and that model performance, price and data retention requirements continue to affect the direction of procurement.

Additional information:The report also mentions that Fable, as a high-end model level for Anthropic, is more oriented towards specific scenes than the universal chat robot market. Anthropic has previously reminded Fable users that the relevant data needs to be kept for 30 days, an arrangement that at one time caused some user dissatisfaction.